As an Amazon Associate I earn from qualifying purchases @AMAZON
Drobo 5D Read/Write Performance
Related: backup, diglloydTools, DiskTester, hard drive, storage, Thunderbolt, Toshiba, video
Testing was on the fast 2.7 GHz MacBook Pro with Retina display via Thunderbolt.
Sequential read/write performance
The Drobo 5D when configured with five 4TB drives is a huge volume, and it simply takes longer than I wanted to run the test (at least 24 hours).
So instead, it was handed an easy test: a 1TB partition out of the ~15TB of capacity. This should allow the Drobo 5D to show its very best performance. Actual performance is likely to be slower as capacity fills up and the slower areas of the drives are used.
Fill-volume performance on 1TB partition
For perspective on the figures that follow, Thunderbolt is capable of up to 1000MB/sec, with realistic real-world figures in the 650 MB/sec range. A single fast hard drive is capable of nearly 180MB/sec or even a little more.
This test was run on a 1TB partition, presumably the fastest part of the ~15TB of available storage. It was the first partition created by Disk Utility (2 partitions created, with regular hard drives the first partition is always the fast one).
Write performance is somewhat better than a single fast hard drive. From the sound of it, the Drobo 5D does a significant amount of head seeking in order to write parity information.
The dropoff in write performance as the test progresses is not encouraging, nor are the inconsistent up/down spikes. This would not be a good choice for 1080P video capture (dropped frames).
Read performance is more than adequate for many tasks and should prove quite satisfactory for most users.
It appears that the Drobo would make a poor choice for a scratch drive or similar (slow writes), but that as part of workflow in which reads dominate it would perform at a very satisfactory level. Generally speaking, reads happen many times, and writes happen once (for image files, video, etc).
MPG testing always uses the command line version of DiskTester as shown below.
- Write performance is somewhat faster than a single fast hard drive (remember, the Drobo 5D is using five fast 4TB drives).
- Read performance is less than one might hope for. Still, it is more than adequate for many purposes, and faster than a SATA II 3 Gbps can perform.
llcR:DIGLLOYD lloyd$ disktester run-sequential -i 10 -t 16G d5d DiskTester 2.2 64-bit, diglloydTools 2.2.0, 2012-12-15 18:05 Copyright 2006-2012 DIGLLOYD INC. All Rights Reserved Use of this software requires a license. See https://macperformanceguide.com/Software-License.html OS X 10.8.2, 8 CPU cores, 16384MB memory Thursday, January 31, 2013 19:29:20 Pacific Standard Time disktester run-sequential -i 10 -t 16G d5d Allocating maximum size contiguous file on "d5d" (16.0TB)...16.0TB (99.9% of volume size) TEMP FILE: /Volumes/d5d/disktester-test/DiskTester-Temporary-Test-File Using test size of 16GB, 4MB at a time at start (0%), within a 16.0TB test file. Iteration 1: writing...203MB/sec, reading...292MB/sec Iteration 2: writing...213MB/sec, reading...374MB/sec Iteration 3: writing...227MB/sec, reading...373MB/sec Iteration 4: writing...233MB/sec, reading...364MB/sec Iteration 5: writing...229MB/sec, reading...368MB/sec Iteration 6: writing...232MB/sec, reading...361MB/sec Iteration 7: writing...234MB/sec, reading...362MB/sec Iteration 8: writing...233MB/sec, reading...367MB/sec Iteration 9: writing...233MB/sec, reading...370MB/sec Iteration 10: writing...232MB/sec, reading...366MB/sec Thursday, January 31, 2013 19:29:20 Pacific Standard Time, volume "d5d" (16.0TB) ----------------- Averages for "d5d" (16GB/4MB, 10 iterations) ----------------- Iteration Write MB/sec Read MB/sec 1 203 292 2 213 374 3 227 373 4 233 364 5 229 368 6 232 361 7 234 362 8 233 367 9 233 370 10 232 366 Slowest 203 292 Fastest 234 374 Average 227 360 Median 232 367 Range 31.3 81.3